Traditional methods for determining critical parameters are often influenced by human factors. This research introduces a physics-inspired adaptive reinforcement learning framework that enables agents to autonomously interact with physical environments, simultaneously identifying both the critical temperature and various types of critical exponents in the Ising model with precision. Interestingly, our algorithm exhibits search behavior reminiscent of phase transitions, efficiently converging to target parameters regardless of initial conditions. Experimental results demonstrate that this method significantly outperforms traditional approaches, particularly in environments with strong perturbations. This study not only incorporates physical concepts into machine learning to enhance algorithm interpretability but also establishes a new paradigm for scientific exploration, transitioning from manual analysis to autonomous AI discovery.
Abstract In recent years scale invariant scattering theory provided the first exact access to the magnetic critical properties of two-dimensional statistical systems with quenched disorder. We show how the theory extends to the overlap variables entering the characterization of spin glass properties. The resulting exact fixed point equations yield both the magnetic and, for the first time, the spin glass renormalization group fixed points. For the case of the random bond Ising model, on which we focus, the spin glass subspace of solutions is found to contain a line of fixed points. We discuss the implications of the results for Ising spin glass criticality and compare with the available numerical results.
Abstract Dynamic space packing (DSP) is a random process with sequential addition and removal of identical objects into space. In the lattice version, objects are particles occupying single lattice sites, and adding a particle to a lattice site leads to the removal of particles on neighboring sites. We show that the model is solvable and determine the steady-state occupancy, correlation functions, desorption probabilities, and other statistical features for the DSP of hyper-cubic lattices. We also solve a continuous DSP of balls into R d .
Abstract Planar run-and-tumble walks with orthogonal directions of motion are considered. After formulating the problem with generic transition probabilities among the orientational states, we focus on the symmetric case, giving general expressions of the probability distribution function (in the Laplace–Fourier domain), the mean-square displacement and the effective diffusion constant in terms of transition rate parameters. As case studies we treat and discuss two classes of motion, alternate/forward and isotropic/backward, obtaining, when possible, analytic expressions of probability distribution functions in the space-time domain. We discuss at the end also the case of cyclic motion. Reduced (enhanced) effective diffusivity, with respect to the standard 2D active motion, is observed in the cyclic and backward (forward) cases.
We considered a long-range Ising model under Glauber dynamics and calculated the difference from the mean-field approximation in a finite-size system using perturbation theory. To deal with the BBGKY hierarchy, we assumed that certain types of extensive properties have a Gaussian distribution, which turned out to be equivalent to the Kirkwood superposition approximation within the range of first-order perturbation. After several calculations, ordinary differential equations that describe the time development of a two-body correlation were derived. This discussion is the generalization of our previous study which developed a similar consideration on the infinite-range Ising model. The results of the calculation fit those of the numerical simulations for the case in which the decay of the interaction was sufficiently slow; however, they exhibited different behaviors when the decay became rapid.
Abstract A 1D model of interacting particles moving over a periodic substrate and in a position dependent temperature profile is considered. When the substrate and the temperature profile are spatially asymmetric a centre-of-mass velocity develops, corresponding to a directed transport of the chain. This autonomous system can thus transform heath currents into motion. The model parameters can be tuned such that the particles exhibit a crossover from an ordered configuration on the substrate to a disordered one, the maximal motor effect being reached in such a disordered phase. In this case the manybody motor outperforms the single motor system, showing the great importance of collective effects in microscopic thermal devices. Such collective effects represent thus a free resource that can be exploited to enhance the dynamic and thermodynamic performances in microscopic machines.
Abstract The characteristic function for heat fluctuations in a nonequilibrium system is characterised by a large deviation function whose symmetry gives rise to a fluctuation theorem. In equilibrium the large deviation function vanishes and the heat fluctuations are bounded. Here we consider the characteristic function for heat fluctuations in equilibrium, constituting a sub-leading correction to the large deviation behaviour. Modelling the system by an oscillator coupled to an explicit multi-oscillator heat reservoir we evaluate the characteristic function.
Abstract This article presents a review on the extension of the idea of linear response theory (LRT) that is well known in the statistical mechanics of systems near thermal equilibrium state, so as to be applied to the statistics in turbulence. The idea has been applied to the statistics in the inertial subrange of turbulence in an unbounded fluid domain. Recently, the idea was extended to one-point statistics in the inertial sublayer of a wall bounded turbulence. A similarity between the energy flux, from large to small scales, in homogeneous isotropic turbulence and momentum transfer in the wall-normal direction in wall bounded turbulence plays a key role in the extension. This article presents also a discussion on further extension of the scope of LRT to the statistics of single-particle diffusion in turbulence.
Abstract Equilibrium and nonequilibrium systems exhibit power-law singularities close to their critical and bifurcation points respectively. A recent study has shown that biochemical nonequilibrium models with positive feedback belong to the universality class of the mean-field Ising model. Through a mapping between the two systems, effective thermodynamic quantities like temperature, magnetic field and order parameter can be expressed in terms of biochemical parameters. In this paper, we demonstrate the equivalence using a simple deterministic approach. As an illustration we consider a model of population dynamics exhibiting the Allee effect for which we determine the exact phase diagram. We further consider a two-variable model of positive feedback, the genetic toggle, and discuss the conditions under which the model belongs to the mean-field Ising universality class. In the biochemical models, the supercritical pitchfork bifurcation point serves as the critical point. The dynamical behaviour predicted by the two models is in qualitative agreement with experimental observations and opens up the possibility of exploring critical point phenomena in laboratory populations and synthetic biological circuits.
Abstract We investigate the critical behavior of disordered systems transversely driven at a uniform and steady velocity. An intuitive argument predicts that the long-distance physics of D -dimensional driven disordered systems at zero temperature is the same as that of the corresponding -dimensional pure systems in thermal equilibrium. This result is analogous to the well-known dimensional reduction property in thermal equilibrium, which states the equivalence between D -dimensional disordered systems and -dimensional pure systems. To clarify the condition that the dimensional reduction holds, we perform the functional renormalization group (FRG) analysis of elastic manifolds transversely driven in random media. We argue that the nonanalytic behavior in the second cumulant of the renormalized disorder leads to the breakdown of the dimensional reduction. We further found that the roughness exponent is equal to the dimensional reduction value for the single component case, but it is not for the multi-component cases.
We present extensive numerical simulations of a family of non-equilibrium Potts models with absorbing states that allows for a variety of scenarios, depending on the number of spin states and the range of the spin-spin interactions. These scenarios encompass a voter critical point, a discontinuous transition as well as the presence of both a symmetry-breaking phase transition and an absorbing phase transition. While we also investigate standard steady-state quantities, our emphasis is on time-dependent quantities that provide insights into the transient properties of the models.
We discuss how to derive a Langevin equation (LE) in non standard systems, i.e. when the kinetic part of the Hamiltonian is not the usual quadratic function. This generalization allows to consider also cases with negative absolute temperature. We first give some phenomenological arguments suggesting the shape of the viscous drift, replacing the usual linear viscous damping, and its relation with the diffusion coefficient modulating the white noise term. As a second step, we implement a procedure to reconstruct the drift and the diffusion term of the LE from the time-series of the momentum of a heavy particle embedded in a large Hamiltonian system. The results of our reconstruction are in good agreement with the phenomenological arguments. Applying the method to systems with negative temperature, we can observe that also in this case there is a suitable Langevin equation, obtained with a precise protocol, able to reproduce in a proper way the statistical features of the slow variables. In other words, even in this context, systems with negative temperature do not show any pathology.
Exact solutions of interacting random walk models, such as 1D lattice gases, offer precise insight into the origin of nonequilibrium phenomena. Here, we study a model of run-and-tumble particles on a ring lattice interacting via hardcore exclusion. We present the exact solution for one and two particles using a generating function technique. For two particles, the eigenvectors and eigenvalues are explicitly expressed using two parameters reminiscent of Bethe roots, whose numerical values are determined by polynomial equations which we derive. The spectrum depends in a complicated way on the ratio of direction reversal rate to lattice jump rate, $ω$. For both one and two particles, the spectrum consists of separate real bands for large $ω$, which mix and become complex-valued for small $ω$. At exceptional values of $ω$, two or more eigenvalues coalesce such that the Markov matrix is non-diagonalizable. A consequence of this intricate parameter dependence is the appearance of dynamical transitions: non-analytic minima in the longest relaxation times as functions of $ω$ (for a given lattice size). Exceptional points are theoretically and experimentally relevant in, e.g., open quantum systems and multichannel scattering. We propose that the phenomenon should be a ubiquitous feature of classical nonequilibrium models as well, and of relevance to physical observables in this context.
A phenomenological construction of quantum Langevin equations, based on the physical criteria of (i) the canonical equal-time commutators, (ii) the Kubo formula, (iii) the virial theorem and (iv) the quantum fluctuation-dissipation theorem is presented. The case of a single harmonic oscillator coupled to a large external bath is analysed in detail. This allows to distinguish a markovian semi-classical approach, due to Bedeaux and Mazur, from a non-markovian full quantum approach, due to to Ford, Kac and Mazur. The quantum-fluctuation-dissipation theorem is seen to be incompatible with a markovian dynamics. Possible applications to the quantum spherical model are discussed.